I was looking at Tim Escobedos "Photon Filter" today
http://www.geocities.com/tpe123/folkurban/fuzz/snippets (scroll down the page)
He says it uses an LED as a variable resistor to ground?! :?
If this can be done could one use an envelope follower to modulate the LED for a simple autowah?
Y
yes.... im working on a filter circuit right now that uses one in this manner.... I may end up switcthing to an optoisolator though
led's are used all over the place as voltage controlled resistors.
Being used as variable resisitors LEDs can operate with very low-voltage signals without considerable non-linearity.
So please check if the AC voltage on the LED doesn't exceed 20 to 50 millivolts. Otherwise you'll have unwilling distortion.
Certainly possible. I first saw the idea from John Simonton's designs. Not a good idea if you demand the lowest distorion... kind of hack quality. But then again, so are 90% of the commercial designs out there. :wink:
This is very interesting!
How does it work? The resistance varies acording to the DC voltage across the LED? It's not supposed to be fairly constant that DC votage?
Good luck!
Miguel
:shock:
So does it get more resistant as more voltage is applied? Or does it get less resistant?
I suppose it gets less resistant, because with more voltage the led will draw more current, and more current means less resistance.. am I right?
Luck
Miguel
I think it is just that way. I have had an old magazine, maybe 1986 and it was in Popular Electronics if I remeber and there was some info. Maybe I could find the mag some day, if somebody is interested... Some discussion has been in this forum, and R.G. mentioned something of the fact that both the signal and control current is small...
Thanks for the tip to use LEDs, I have only few times tried with silicon signal diodes. I did a drawing from memory last weekend about interesting audio level control/attenuator. I tested it with some signal from my cassette player or cd, and it worked quite nice. What I did not try was how it distorts when level is raised, and don´t know if you could put in germaniums and leds.
http://img.photobucket.com/albums/v411/Nasse/diddex.jpg
think of it like this.... this is kind of backwards, but will help you understand how it works.
if you have a voltage with some resistance to one side of the led, and the other side goes to ground the led will always do what ever it can to make sure that the voltage accross it is constant unless it goes below the turn on voltage. The control voltage will affect the LED (or any diode for that matter) in two different ways.
if the control voltage changes the diode resistance will change
if the impedance of the control voltage changes the diode resistance will change.
more voltage=less resistance
more impedance=more resistance (but possibly more distortion)
I think I get it now, thanks for the explanations
Now I got another circuit bit for making designs, this forum keeps rocking!
Luck
Miguel
IMHO linearised FET (1/2 of drain voltage on the gate) is not too much more complicated. At the same time it can handle up to 200 mV RMS with distortion low enough for guitar purposes.
Sure, but is nice to have other options too, and I can get Leds much cheaper than fets!
Luck
Miguel
Re: diodes as variable resistors.
As we all know, diodes are non-linear - big time! The idea of using them for variable (and implicitly linear) resistors seems odd. However, it is possible under certain circumstances.
Those circumstances all involve small signals. Think of what we know about diodes. At quite low voltages, diodes do not conduct; that is, they act like a high resistance. At some larger voltage, the diodes conduct, and hence act like low resistances. The whole trick is to make a bias voltage/current flow in the diode so that for tiny, tiny perturbations from that bias point, the diode acts more or less like a linear resistor.
I can do this lecture more easily with drawings, but I'll try typing.
Consider a resistor. If you put a voltage across it, it lets a current flow in response to I = V/R. If we put in a number of voltages, we can compute a number of resulting currents. We can then make a graph of the currents resulting from changing the voltages. Imagine a square graph, with current on the vertical axis and voltage on the horizontal axis. We know that with no voltage there is no current, so there is a point at 0A, 0V. At V=1, the current is 1/R. At V=2, I=2/R. Pretty quickly we find that the resistor graph is a straight line from the origin at 0,0 up and to the right at a slope of 1/R.
Now lets do that with a diode. The diode will start at 0,0 too, as it produces no voltage/current (in the dark at least) As we increase the voltage to 0.1V, 0.2V, etc. only the most miniscule currents flow. We can compute the "resistance" of the diode between 0.2V and 0.1V by subtracting the currents at those two points and the voltages, then taking the ratio of the two, as R = (V2-V1)/(I2-I1). We know that the resistance is high at these quite small voltages.
If we get up at (for silicon) 0.7 to 0.8V, we will find currents that go from perhaps 0.1A to 10A, making for a "resistance" of R = 0.1V/9.9A = 10 milliohm.
In the middle between these two, the diode resistance has changed from fractional megohms to milliohms.
The catch is that we would like the diode to act like a resistor. And we can do that if we restrict the applied signal voltage across the diode to, say, 25mv. A diode with 400mv +/- 25mv doesn't change resistance very much between 375mv and 425 mv, so the distortion is not very noticeable. The smaller the signal variation, the smaller the resistance change and the smaller the distortion. A 1mv signal (common enough in RF work) would be 399 to 401mv, and the resistance of the diode is pretty darn linear.
You have to keep the signals **small** for diodes of any stripe to not distort. What counts is the degree of curviness of the conduction knee of the diode. LEDs are a bit more softly rounded than silicon, so they have a bigger signal carrying capability than silicon in varible resistance use.
Historically, the Magnatone amps used voltage variable resistors called MOVs to provide variable resistances for phase shift networks. These things went from maybe 100K to 10K over a 70V range, so even volt-size signals could be used on them.
RG, does this means the control voltage has to be in a very well calculated range?
I see more use for this as a shunt switch than a means for modulating signals, at least can be used in a non-audio path where distortion may be less important
Good Luck!
Miguel
Yes, tis very problematic in an envelope follower, what I saved not buying FETs/LDRs, I have spent on trimpots. Its getting there though, once I have something that works/is useable, i'll post it here.
Thanks for your replies :)
QuoteI see more use for this as a shunt switch than a means for modulating signals, at least can be used in a non-audio path where distortion may be less important
If you're doing a shunt switch, do it the easy way - use a bipolar transistor.
Hook up an NPN with emitter to ground, base through resistor to control voltage, and collector through a big capacitor to the signal to be shunted. Current into base is signal shunted, base held to ground is no shunt. It's good up to about a diode drop for signals with base at ground. You can extend to higher signal levels by pulling the base below ground.
Diode shunt switches have big offset voltage problems.
What about me - I do prefer using CA3080 or LM13700 for modulators/controllable gain and filter stages, etc. They are highly effective, linear, low consumption relatively to LED-LDR pairs, handle "high'-voltage signal and can not only control signal but amplify it at the same time. They are also very reliable and "undemanding".
I'll stick to bipolars then (also cheaper than leds). Is good to know how this things could work anyway.
I've designed a little tremolo using a bipolar transistor shunting technique, but doesnt sound as good as I wished. Should I post a sound sample in case someone finds it interesting?
Thanks for the help
Miguel
bioroids,
BJT switch is good only in two positions - fully ON or fully OFF. It works poorly (as a key) when it is in its "active" zone between ON and OFF.
Since your LF tremolo oscillator generates sine or triangle wave, your BJT works in-between ON and OFF states during some parts of every LF cycle. Hence your sound is distorted quite audibly.
Quote from: DDDWhat about me - I do prefer using CA3080 or LM13700 for modulators/controllable gain and filter stages, etc. They are highly effective, linear, low consumption relatively to LED-LDR pairs, handle "high'-voltage signal and can not only control signal but amplify it at the same time. They are also very reliable and "undemanding".
All of the above is true, but on the other hand they are more expensive than LEDs, BJTs, FETs, Vactrols etc. and are more difficult to get hold of :)
DDD, you're right about that, it's kinda dirty sounding and I think the modulation gets square in spite of the lfo waveform (taken from the EA tremolo, roughly sine I think)
I posted a few samples on my site http://ar.geocities.com/bioroids
if you want to hear a cheap designed tremolo
lightningfingers, i think the CA3080 is available from Small Bear, and not too expensive, though is easier to get hold of fets or bjts
regards
Miguel
bioroids,
since every modulator is actually a signals' multiplier, you have a lot of frequency products on its output, so every harmonic of your distorted pseudo-sine LFO generates numerous audible products.
In my opinion it's possible to build good distortion box using cheap BJTs and diodes but it's virtually impossible to build decent time effects or modulators without more expensive parts.
Do you mean, DDD, the difference and sum of the signal frequencies and the LFO harmonics appear at the output? interesting
This does not happen with fet or ota based tremolos? or does it depend on the modulation waveform?
A few months ago I used this (with 4066 switches) for an 'aliasing simulator', it worked nice (I'm not sure what were the frequencies comming out of the circuit) but I was using a high freq oscillator (HFO?)
Regards
Miguel
Quote from: lightningfingersAll of the above is true, but on the other hand they are more expensive than LEDs, BJTs, FETs, Vactrols etc. and are more difficult to get hold of
Just about every electronic component supplier in the UK sells the CA3080 & LM13700. When you tot up the additonal cost of LEDs & LDRs/JFETs/ CMOS chips for matched MOSFETs, plus the extra board space for these components and extra current consumption if you go down the LED/LDR route, or the extra work of matching JFETs, then OTAs look pretty cheap to me (not that they're expensive). Plus you'll get a more linear reproduction of your LFO waveform. :wink:
Is someone using the NJM13600D otas? they are a little cheaper
I've never messed with this things before, but they seem to be nice little ICs. Luckily I have a few (for a Small Stone) so I'm gonna try a few things these days (when I finish moving!)
Regards
Miguel